🚀 go-pugleaf

In article <425@ivax.doc.ic.ac.uk> whw@doc.ic.ac.uk (Victor Wu) writes:
>	I've recently bought a Wyse-700 monitor for my AT-clone and trying to
>port the bitblter for the X windowing system (from Brown University) to run on
>it. However, I found it is very difficult to work with this particular
>monitor. The reason being that when it is at its highest resolution (ie. 1280
>X 800 pixels black and white), the frame buffer is divided into two banks of
>memory. One for the even scanlines and the other for the odd scanlines.

I haven't looked at the Brown code lately (and it's hard to find it when
you haven't told me which version of X you're using or which device it
was written for!), but in general it's easy to handle this by blt-ing
the odd and even lines separately.

Create dummy bitmap structures for the source and destination bitmap
operands with the bytes-per-scan-line values doubled.  (Think of this as
creating new bitmaps with all the original even lines on the left side
and the original odd lines on the right side.)  Now you can perform most
bitblts by calling the regular bitblt function twice with slightly
munged dx, dy, w, h, sx, sy operands (swapping the frame buffer bank in
between).

There is one pathological case where the source and destination bitmaps
are both areas on the frame buffer and the LSBs of the source and
destination y coordinates differ (e.g. dy is even and sy is odd).  This
has be done with a temporary memory bitmap (which only has to be half
the size of the bitblt area).
-- 
David DiGiacomo, Sun Microsystems, Mt. View, CA sun!david david@sun.com